Luca Steeb

We racked up $13.3M in 48 hours after going viral on X, in annualized pledged token spend. Here's the full breakdown of the results + answers to some questions about our 0% markup alternative to OpenRouter. 1. The result, what ~600K impressions on X got us: > ~$1.11M/mo in pledged token spend > 250B+ tokens/month in estimated volume > 14 customers spending $50K+/mo on tokens > 90+ customers doing $2K+/mo, carrying 94% of all pledged dollars > 400+ customers doing $200+/mo > 903 applications total, from 101 countries 2. Key findings on token spend habits: > Where devs route tokens from: 61% OpenRouter, 57% direct from labs (Anthropic / OpenAI / Google), 31% smaller routers. > What they spend on: 47% dev tools + coding assistants, 34% AI agents/automation, 3% data extraction. > Throughput: 5% need 300+ req/min, 20% need 60+ req/min (exactly why the labs' rate limits suck) > Top Regions: 19% USA, 18% India, 4% UK, rest spread across 98+ countries 3. Questions, comments, and concerns > How do you guys make money? This was by far the most asked question. The answer is simple, we buy large volume upfront from providers at below list cost and then sell it at list price w/o any markup. After absorbing the transaction fees we come out around 1.5% model blended margin. At $1.11M/mo of pledged spend, that's about $16K/mo take home. Not bad for our first month of existence. We expect to grow this by at least 50x within the next 12 mo's. >What's ur moat ? What if OpenRouter does this tmr? They're more than welcome to! Matter of fact they should. We're not going to pretend there's some insane technical moat. Routers are a commodity service and switching cost is zero. That cuts both ways: it's how we take their customers with a one-line config change. But they can't match us without deleting the revenue Stripe just paid $7B for. Our job is to grab as much market share as possible while it's open, and 0% markup is the easiest sale there is. > Okay so its race to the bottom and ur gonna die out eventually ? No, also routing is just the beginning! Ultimately, our vision is a marketplace where inference workloads and commit volume can be bid on. Imagine a customer putting up a listing of their projected monthly spend, say 50B Kimi-K3 tokens, and inference providers like Fireworks, DeepInfra, competitively bid on the contract, so the customer gets the fairest possible price. Flip side, where there's tons of demand for dedicated capacity, a provider lists their stock instead, ex: "we can serve 50B GLM 5.3 tokens/mo" and buyers bid on that. Then say you bought $100K in commits and your plans changed, you don't need them anymore, or they're worth more now than when you bought. You just resell them back to the market. Like with all marketplaces though, liquidity is the main concern. Who comes first, sellers or buyers? Our goal with Straitly is to gather that liquidity, and the 0% markup ofc helps that tremendously, every customer we onboard is a buyer already on the book. Once we attain a reasonable amount of liquidity on both sides, we'll bring a whole new dimension to this market. ps. We'd love to meet everyone's demand but we're currently maxed out on capacity, in the process of acquiring more as I write this, so we're onboarding in batches, if you have not heard from us, do not worry we are working hard to get your share. Apply here: > Customers, 0% markup, 30% off your first $10K, $100 upfront: straitly.ai/apply/customers > Providers, if you can serve tokens at scale we want you on: straitly.ai/apply/providers Also: we're hiring engineers, if ur in SF, lets chat.

JUST IN: Stripe acquires AI model marketplace OpenRouter for more than 5x its valuation three months ago

🫡